An optical integrated circuit comprising a semiconductor substrate, and optical function elements, e.g., a laser source, an optical modulator, an optical detector and an optical branching filter, and optical waveguides integrated on the substrate has been developed as an element for optical communication.
In the optical multiplexer and demultiplexer, distributed feedback laser, etc. of the above-described optical elements, a diffraction grating is utilized. Thus, the production of such optical integrated circuits needs a technique which permits the formation of a diffraction grating at a limited area on the semiconductor substrate.
Photolithography is now used in the production of function elements. Of course, a diffraction grating can be formed by a similar technique.
In forming a diffraction grating on a photoresist, however, there cannot be employed the usual method that comprises placing a photomask having a pattern to be formed, on a substrate and, thereafter, exposing it to light, because it is not possible to form a fine grating pattern in the photomask.
In the formation of such diffraction gratings, therefore, the holographic exposure method has been used.
The schematic diagram of a conventional exposure apparatus which is used to form a diffraction grating according to the holographic exposure method is shown in FIG. 1.
Referring to FIG. 1, a beam of light leaving a laser 1 reaches a beam splitter 2 where it is split into two light fluxes. Each light flux is then converted into a parallel light flux having a greater beam diameter by means of the corresponding collimator 3 or 4. The thus-enlarged light flux is reflected by a mirror 5 or 6 and irradiated on the surface of a semiconductor substrate 7 coated with a photoresist 8.
Since the coherent light from the laser 1 is split into two light fluxes and then incident on the surface of the photoresist 8 at a pre-fixed angle relative to each other, the photoresist 8 is exposed to an exposure energy changing in a sine wave form along the line where a plane containing the two light fluxes and the photoresist surface intersect. Upon appropriate development of the thus-exposed photoresist, a part of the photoresist remains unremoved in a grating form, resulting in the formation of a diffraction grating. The period of the grating can be changed appropriately and optionally by changing the angle at which the light is incident on the photoresist 8.
In accordance with this method, it is possible to expose the photoresist in an interference pattern of 1 .mu.m or less because there cannot be formed a photomask having a diffraction grating pattern of a submicron cycle.
The conventional two light flux interference method, however, has the disadvantage that since beams having a large diameter are made to interfere, it is possible to form a uniform diffraction grating over a large area, but a diffraction grating of the desired size cannot be formed within a limited area. Thus, in accordance with the conventional holographic exposure method, it is not possible to produce optical integrated circuits having a diffraction grating.
The third method is an electron beam exposure method which utilizes electron beams in the formation of diffraction gratings. Since it is possible to control the trace of electron beam with an accuracy of B 1 .mu.m, there can be formed a diffraction grating of a submicron period. This method, however, needs a large-sized apparatus, which will lead to an increase in production costs.
Diffraction gratings are widely used in spectrometers, or as optical branching devices for optical communication because of their high wavelength selectivity and resolving power.
These diffraction gratings can be prepared by a mechanical process, or the above described holographic exposure process in which interference fringes due to two beams are formed on a photoresist.
The mechanical process is a method in which a number of equidistant parallel lines are ruled on a substrate by the use of a diamond cutter. These lines can be formed in any desired form. This method, however, has disadvantages in that much complicated labor and long working times are needed because it is necessary to rule from 1,000 to 2,000 lines per millimeter, one by one. Therefore, the production costs are undesirably increased.
In accordance with the holographic exposure process, a diffraction grating is formed by a photographic technique; i.e., a substrate, e.g., glass, coated with a photoresist is irradiated with two coherent beams in such a manner that the beams form a suitable angle relative to each other, and a diffraction grating corresponding to the resulting interference pattern is formed. Upon development of the photoresist, unexposed areas (in the case of negative type photoresists) or exposed areas (in the case of positive type photoresists) are removed, leaving a number of parallel grooves. Vacuum-deposition of a suitable metal, e.g., aluminum, on the grooves provides a diffraction grating of the reflection type, comprising a number of equidistant parallel lines.
This diffraction grating is also called a "holographic grating". The cross section of the diffraction grating is in a sine wave form as illustrated in FIG. 2. In such reflection type diffraction gratings, if the cross section is in a sine wave form, diffraction of high efficiency cannot always be expected. This is because in the case of such sinusoidal gratings, even if light having any wavelength is incident on the grating at any incident angle, there exist an angle of diffraction and an order of diffraction meeting the requirements for Braff diffraction in a broad sense and, therefore, diffracted light is scattered in many directions.
On the other hand, the use of a diffraction grating having a cross section as shown in FIG. 3, i.e., comprising a number of parallel triangular projections having a long slanting surface BA and a short slanting surface AC, increases diffraction efficiency. Diffraction grating having cross sections as shown in FIG. 3 are called "blazed diffraction gratings" because the surface BA is slanted. The angle (.alpha.) between the normal PR of the grating surface and the normal QR of the slanting surface BA is called a "blaze angle"; i.e., &lt;ABC=.alpha..
In the blazed diffraction grating of FIG. 3, the energy of light which is incident normally on the slanting surface AB and diffracted therefrom at the same angle as above is much greater than that of light which is diffracted in a direction corresponding to another order of diffraction. The relation between the wavelength of the light (blaze wavelength), .lambda..sub.B, and the blaze angle is represented by the formula: EQU d sin .alpha.=.lambda..sub.B ( 1)
where d is a grating constant.
In addition to a beam of light having the blaze wavelength, there is another beam of light which is diffracted particularly strongly. This beam of light is such that the wavelength and the angle of incidence satisfy the Bragg condition, and the direction of incidence and the direction of diffraction are symmetrical in relation to the normal QR.
In this way, the blazed diffraction grating provides a high diffraction efficiency for a beam of light having a specific wavelength and a specific angle of incidence.
In preparing a diffraction grating according to the holographic exposure method, only one exposure produces a diffraction grating having a cross section of the sine wave form as shown in FIG. 2, whereas when exposure is applied twice, there will be obtained a diffraction grating similar to the blazed diffraction grating.
In general, a periodic function where the period is d can be expanded into a Fourier series.
The cross section of the blazed diffraction grating can be made to correspond to the graph shown in FIG. 4. This is a periodic function and, therefore, can be expanded into a Fourier series as follows: ##EQU1## where EQU k=(2.pi.)/d. (3)
Since f (x) is an odd function in relation to the origin 0, it can be expanded as a sine function.
The coordinates of Point A are determined by a blaze angle, .alpha., an angle A, and a period, d. In a case in which the angle A is 90.degree., for simplicity, the coordinates of Point A are as follows: ##EQU2##
Fourier coefficient bn is determined by the following equation: ##EQU3##
This can be integrated as follows: ##EQU4##
This Fourier coefficient is nearly in reverse proportion to the square of n. Therefore, it converges uniformly at a relatively high speed. In particular, when the terms are n=1 and n=2 are taken and the subsequent terms (n=3, . . . ) are dropped out, the resulting function is believed to appropriately represent the wave form. Although acute angles A, B, C, . . . as shown in FIG. 4 cannot be represented by the terms n32 1 and n=2, the function EQU f.sub.2 (x)=b.sub.1 sin kx+b.sub.2 sin 2 kx (6)
appropriately represents the wave form of the blazed diffraction grating.
In FIG. 5, (a) represents b.sub.2 sin kx, (b), b.sub.2 sin 2kx, and (c), (b.sub.1 sin kx+b.sub.2 sin 2kx). (c) is f.sub.2 (x) of the equation (6), and is very similar to the function of the surface of the blazed grating.
Thus, when an interference fringe having a wave number of k and an interference fringe having a wave number of 2k are exposed in a double form with suitable weights (amplitudes) b.sub.1 and b.sub.2, there can be prepared a diffraction grating similar to the blazed diffraction grating.
This technique has already been proposed and is well known. This technique, however, is difficult to employ. The difficulty is that the origins X=0 of the two functions sin kx and sin 2kx must be in agreement with each other. If the positioning is not complete and, as a result, there is formed a phase gap .phi., the resulting function is represented as follows: EQU .phi.(x)=b.sub.1 sin kx+b.sub.2 sin (2kx+.phi.) (7)
This function cannot represent the graph as shown in FIG. 5(c).
It is required for the origin X=0 to coincide in both the functions with much higher accuracy than the grating distance d. Since this positioning is very difficult, a method of forming diffraction gratings by double exposure has not yet been put to practical use.
The optical wavelength division multiplex communication system (WDM) transmitting simultaneously a number of light waves having different wavelengths by means of one optical fiber has been extensively studied because of its potentiality for a large amount of communication. An optical branching filter element is a device which is used to take out light having a specific wavelength of a multiple optical signal. Thus, the optical branching filter element is one of the devices which play a significant role in the optical wavelength division multiplex communication system.
Optical branching filter elements which have now been almost put to practical use include a diffraction grating and an interference filter, which are fabricated in a three-dimensional structure.
FIG. 6 shows a cross section of one example of the known diffraction grating type optical branching filter elements. This optical branching element comprises a diffraction grating 31 accommodated in a box type casing 30. When light from an optical fiber 32, containing light having a wavelength .lambda..sub.1 and light having a wavelength .lambda..sub.2 enters the optical branching filter element, the light having a wavelength .lambda..sub.1 and the light having a wavelength .lambda..sub.2 are diffracted in different directions by the diffraction grating 31, whereby the light from the optical fiber 32 is branch-filtered. The light having a wavelength .lambda..sub.1 and the light having a wavelength .lambda..sub.2 can be taken out through light outlet apertures 33 and 34, respectively, which are provided at locations corresponding to the given diffraction angles.
FIG. 7 shows a cross section of one example of the known multi-layer membrane filter type optical branching filter elements. This is a three dimensional optical branching filter element comprising an interference filter 35 accomodated in a box type casing 30. When light from an optical fibre 32 enters the casing 30, it is divided into light having a wavelength .lambda..sub.1 and light having a wavelength .lambda..sub.2 by the interference filter 35. This multi-layer membrane 35 comprises a number of dielectric thin films having different refractive indexes which are superposed on each other, and lights reflected from the boundary surfaces interfere with each other. The interference filter 35 can reflect almost 100% of light having a specific wavelength and conversely, can transmit almost 100% of light having a specific wavelength.
These conventional optical branching filter elements, however, have disadvantages in that they are in a three dimensional structure and are large sized elements. Thus, small sized elements have been desired.
For this reason, two dimensional wave guide type optical branching filter elements have been proposed. These optical branching filter elements have received increasing attention because of their small size and stability, and have been extensively studied.
FIG. 8 is a perspective view of a known wave guide type optical branching filter element. In a waveguide type optical branching filter element 36, a diffraction grating 39 is formed by ruling periodic lines on a light sensitive material 39 provided on a substrate 37 by a photolithographic technique, for example. A two dimensional wave guide 40 is provided in the structure which extends through the diffraction grating 39, or is bent therein. When a combination of lights having wavelengths .lambda..sub.1, .lambda..sub.2, and .lambda..sub.3 is introduced, only the light having a wavelength .lambda..sub.2, satisfying the Bragg condition in relation to the diffraction grating is diffracted, and the remaining lights are allowed to travel straight therethrough. In the wave guide type optical branching filter element, the wave guide and the diffraction grating are on the same plane and in a two dimensional arrangement. Therefore, this type of branching filter element can be reduced in size.
A diffraction grating can be formed in a plane containing a wave guide by techniques such as a holographic exposure method and a light beam scanning method.
As discussed previously with respect to FIG. 1, two beams (a) and (b) form an interference fringe on the surface of the substrate since they are coherent laser beams. The period d of the interference fringe is given by the equation: ##EQU5## Upon development of the above exposed light sensitive material, there is formed a diffraction grating having a sinusoidal cross section.
FIG. 9 shows a schematic diagram of an optical system for the light beam scanning method. A laser beam 45 is focused by means of a lens 46 and scanned on the light sensitive material 38 so that it draws parallel grating lines 47. By scanning the laser beam, the parallel grating lines are ruled one by one to form a diffraction grating. The above described holographic exposure two beam interference and light beam scanning methods are known as optical methods of preparing a diffraction grating.
Although the waveguide type optical branching filter element 36 as shown in FIG. 8 can be prepared by the above described optical methods and is a promising element, the wavelength of light to be branch filtered is fixed. Since the period d of the diffraction grating is fixed, a wavelength satisfying the Bragg diffraction condition is previously determined.
It has thus been desired to develop optical branching filter elements which enable one to freely choose the wavelength to be branch filtered. If the wavelength can be freely chosen, it is not necessary to prepare a variety of diffraction gratings depending on wavelengths. Moreover, if the choice of wavelength can be conducted immediately, it is possible to provide the diffraction grating with an optical switching function.